Date of Award

Spring 2022

Project Type


College or School



Decision Sciences

Departments (Collect)

Health Data Science

Program or Major

Health Data Science

Degree Name

Master of Science

First Advisor

Dr. Joanna Gyory

Second Advisor

Dr. Semra Aytur


The global pandemic that began in the United States in early 2020 continues to be a topic of controversy. The added aspect of affect polarization in the country’s political realm may have exacerbated the effects of COVID-19. In their published article in Nature Human Behaviour, Gollwitzer et. al. found that it was possible to link voting partisanship, physical distancing, and COVID-19 outcomes showing that a county’s partisanship might be used to predict the degree to which that county would socially distance and then, therefore, the rate of cases and fatalities in that error on a lagged timescale. This researcher attempted to replicate and validate the findings of an analysis conducted in the earliest months of the pandemic using approximately the same variables, models, and covariates, but over a longer span of time in the pandemic.

Three possible mediator variables (physical distancing data, mask mandate data, and online sentiment data) were gathered and tested for usability in the main mediation analysis. Preliminary analysis of the data gathered did not support the assertion of sentiment or masking data would be useful to the mediation analysis due to insufficient data. Though the distancing data was significantly linked to partisanship to become a proxy, mixed models showed that pandemic dates after the period of the original analysis could not support physical distancing as a mediator for partisanship. Only the segment of the final dataset which matched the dates of the original work were processed through the same mediation analysis in STATA. Significant effects of partisanship on case growth rates were discovered, but not to the same degree as the original work.